Studia Metrica et Poetica 1.2, 2014, 39–70
Mythological Names and dróttkvætt Formulae II:
Base-Word-Determinant Indexing
Frog*1
Abstract: This article explores patterns of language use in oral poetry within a variety
of semantic formula. Such a formula may vary its surface texture in relation to phonic
demands of the metrical environment in which it is realised. This is the second part
of a four-part series based on metrically entangled kennings in Old Norse dróttkvætt
poetry as primary material. Old Norse kennings present a semantic formula of a particular type which is valuable as an example owing to the extremes of textural variation
that it enables. The first part in this series introduced the approach to kennings as
semantic formulae and included an illustrative case study on kennings meaning ‘battle’
realising the last three metrical positions of a dróttkvætt line. This demonstrated that
lexical variation in realising these formulae varied according to functional equivalence
across semantic categories. The present case study advances this discussion through
the examination of the metrical entanglement of the lexicon in realising the semantic
formula. On the one hand, it presents evidence of the associative indexing of lexical items realising a battle-kenning of this particular metric-structural type: certain
kenning base-words exhibit a preferred semantic category of determinant. On the
other hand, it also presents evidence of the associative indexing of lexical items that
are used for realising the metrically required rhyme in a position in the line that is
outside of the semantic formula: certain kenning base-words exhibit co-occurrence
with a particular rhyme-word.
Keywords: oral poetry, variation, formula, skaldic poetry, dróttkvætt, kenning
This is the second part of a four-part discussion that addresses a phenomenon
that I describe as the ‘metrical entanglement’ of language in an oral-poetic
tradition, looking particularly at verbal variation in semantic formulae. This
phenomenon is addressed through semantic formulae of a particular type
called a kenning in Old Norse skaldic poetry. The article builds on a pilot study
of 340 examples of kennings meaning ‘battle’ in the metre known as dróttkvætt
(Frog forthcoming). The four-part discussion explores phenomena revealed by
* Author’s address: Frog, Folklore Studies / Department of Philosophy, History, Culture and
Art Studies, University of Helsinki, P.O. Box 59 (Unioninkatu 38 A), 00014 University of Helsinki, Finland. E-mail: [email protected].
doi: dx.doi.org/10.12697/smp.2014.1.2.03
40
Frog
the pilot study in greater detail through a series of three case studies of different
metric-structural basic types, which will be followed by a final synthesis and
discussion. The data-set for each case study has been considerably expanded
from that of the pilot study, although it should not be considered exhaustive.
Particular attention is given to variation between the use of nouns referring to
weapons and armour (implements of battle) versus names of valkyries and of
the god Odin (mythological agents of battle) in realising the metrically entangled battle-kenning formulae. This attention is both owing to the unambiguous
difference between these as semantic categories as well as a concern for the
degree to which the vocabulary of proper names was fully integrated into the
poetic register and into the strategies for poetic composition.
The first article in this series (Frog 2014a) presented the register-based
approach to oral poetry employed here as well as the framework for approaching
kennings as semantic formulae. It also introduced the metrical entanglement
of semantic formulae in the case of kennings through basic type 12(p)XYy and
provided an initial demonstration of the integration of personal names into the
registral lexicon of dróttkvætt poetry. The present case study turns to the more
frequently attested basic type 1(p)YyXx discussed in a dataset of 80 examples.
Focus in this case study is on evidence that certain words used in realising a
metrically entangled formula may exhibit a) preferred categories of equivalent
lexical items for completing the formula as well as b) conventionally associated
verbal material for completing the metrical line. In order to improve the general
accessibility of this article, the case study is prefaced with a brief review of the
background, terms and approach employed so that it is possible to follow the
argument and discussion without referring back to the first article in this series.
Background and Terms
Metrical entanglement is here used to describe the phenomenon by which the
language of an oral-poetic system becomes bound up with metrical positions
or other metrical parameters. Metrical entanglement occurs along a continuum, on which degrees of fixity are described in terms of crystallisation,1 at
the extreme of which is ‘metrical boundedness’ with the potential for the ‘fossilisation’ of lexical material. Metrical boundedness was a qualifying feature
of Milman Parry’s early definition of the formula in oral poetry as “an expression which is regularly used, under the same metrical conditions, to express
1
Following Siikala 1990 [1984]; on the choice of this term, see also Frog 2014a: 103n.6.
Mythological Names and dróttkvætt Formulae II
41
a particular essential idea” (Parry 1928: 16, emphasis added), a definition
later propagated by his student Albert Lord (1960: 4). Such prefabricated formulaic expressions have thus received tremendous attention with the rise of
Oral-Formulaic Theory.2 The metrical boundedness of such formulae has
sometimes been exaggerated, and Parry’s advocation of metrical fixity in defining a formula (circularly derived from his statistical method for identifying
formulaic language) proved too narrow for general use (e.g. Hainsworth 1968).
A formulaic expression is approached here as a general linguistic phenomenon
that may also take shape in the metrical environment of oral poetry. Following
Alison Wray (2009: 28–34), a linguistic formula is here defined in terms of
morpheme-equivalence: it is characterised by a coherent unit of meaning with
an exclusive entry in the mental lexicon of language users, even if it can be
analysed and appropriately interpreted according to rules of the grammar (cf.
Parry’s “particular essential idea”; cf. also the discussion of emic conceptions
of a ‘word’ in an oral-poetic register in Foley 1999: 67–69 and the definition of
an oral-poetic formula “as an integer of traditional meaning” in Foley & Ramey
2012: 80). This will provide a frame for considering metrically entangled kennings as semantic formulae exhibiting variation at the lexical surface of the text.
The present investigation is developed on a usage-based approach to language and variation, according to which individuals internalise language and
the strategies and associations of conventional language usages through exposure to and participation in cultural practices. This approach is developed
through register theory (e.g. Halliday 1978; Foley 1995; Agha 2007). Following
register theory, language does not function in social practice as a monolithic
ideal abstraction – la langue of Saussure (1967 [1916]) – but instead is constituted of and internalised through multitudinous varieties and sub-varieties
On Oral-Formulaic Theory and the formula, see e.g. the reviews in Foley 1988 and Foley
& Ramey 2011; cf. also its relation to other approaches to oral and performative expression in
Foley 1995. Scholars of skaldic poetry who are less familiar with Oral-Formulaic Theory or
think of it solely in terms of Albert Lord’s Singer of Tales (1960) may find this frame of reference incongruous. However, Oral-Formulaic Theory has developed considerably across the
past half-century. ‘Composition in performance’ and ‘memorisation’ have not been regarded
as mutually exclusive opposites for a long time. Questions of how the resources of a tradition
relate to the entextualisation of e.g. a narrative or communication are complementary to questions of how the same resources may function as mnemonics for producing similar expressions
(e.g. ‘themes’) or for reproducing socially recognisable texts. Within this frame, resources of
the skaldic register are used for the entextualisation of utterances (whether that process is slow
and reflective or in the situation of performance, as in insult exchanges) that become socially
circulating texts, and fluency in the resources of the idiom must have reciprocally provided the
mental equipment necessary to remember and reproduce those socially circulating texts.
2
42
Frog
linked with situations, contexts, users and uses. The term register is a flexible
tool used to designate such varieties. This term can be calibrated in relation to
the research object. When applied to an oral poetry tradition, the linguistic
register is the historically developed and metrically conditioned language as
used in that poetry, ranging from its lexicon and semantics to grammar, phonology and patterns of use. Thus, whereas modern poetry can and does draw on
the full spectrum of linguistic resources available and can produce meanings
in relation to them (cf. Hasan 1989: 90–106) – i.e. all registers are potentially
open to it – an oral-poetic tradition will be characterised by its own register
and meanings are produced in relation to that framework of linguistic behaviour (cf. Foley 1995; 1999: 65–88). Novelty that significantly deviates from that
framework therefore becomes unlikely insofar as it may threaten the success and
effectiveness of communication (Abrahams 1969: 194; cf. Foley 1991). Within
the register of Old Norse skaldic (and related) poetry, the present discussion
will be particularly concerned with the register (or sub-register) of dróttkvætt
poetry. In this tradition, calibrating register in relation to a metre is relevant
because the broader register of skaldic (and related) poetry was conventionally
applied across a number of metres. This broader register was thus metrically
entangled with each of those metres, especially in relation to their relevant
metrical requirements. In other words, within the register, formulae and similar
resources also took shape in relation to individual metres and their particular
conventions. Insofar as the metres were different, dróttkvætt had, for example,
a distinctive and extensive lexicon of formulae interfaced with the metre and
that functioned as resources for composition in that metre but not necessarily
in others (see further Frog 2014a: 107–109). Thus, addressing a ‘register’ of
dróttkvætt calibrates focus onto linguistic resources that are metrically entangled
with that particular metre. Here, focus will be more specifically on the metrical
entanglement of battle-kennings as semantic formulae in dróttkvætt expression.
The dróttkvætt metre was essentially syllabic (with rule-governed flexibility) composed in couplets, with two couplets forming a half-stanza called
a helmingr. Rules of syntax allowed a remarkable scrambling of language
through a helmingr, and two (rarely more) clauses or independent statements
could be interwoven across those four lines. The main conventional constraints
relevant to the present discussion are rhyme and alliteration; syllable weight,
relevant to the first and third case studies in this series, need not be discussed
here. Two lexically stressed syllables in odd lines should alliterate with the
first stressed syllable in the following even line of a couplet. Rhyme was more
variable in practice, but normally the penultimate syllable (but not the following inflectional ending) of each line should rhyme with a preceding syllable;
in odd lines, this should be skothending rhyme, not including the vowel (e.g.
Mythological Names and dróttkvætt Formulae II
43
1b.i below: Gunnr um geira sennu); in even lines, this should be aðalhending
rhyme, including the vowel (e.g. 1a.i below: menn at vápna sennu).3
A kenning is a rhetorical figure that forms a Noun Phrase (NP). In most
registers (poetic or otherwise), kennings are for the most part crystallised formulaic expressions or have become wholly lexicalised, fossilised and idiomatic.
This rhetorical figure is formed of a noun called a base-word (NP1) complemented by a second noun called a determinant (NP2) in the genitive case or
by forming a compound, and these together signify a third, nominal referent
(thus: NP2-GEN NP1 or NP2-NP1 = NP3 when NP3 ≠ NP1 or NP2). For example, the base-word hríð = ‘storm’ can be complimented by the determinant
sverð = ‘sword’ to form a battle-kenning in the following ways: hríð sverðs =
‘storm of the sword’, hríð sverða = ‘storm of swords’ or sverðhríð = ‘swordstorm’. Kennings in skaldic poetry are exceptional because they functioned
generatively: the register was characterised by a rich lexicon of semantically
equivalent terms called heiti (sg. also heiti) which could be interchangeable in
realising a particular kenning as a semantic unit according to “paradigmatic
substitution” (Clunies Ross et al. 2012: lxxi). Thus in the above example, hríð
is a weather-heiti interchangeable with él = ‘squall’, drífa = ‘snowstorm’, hregg =
‘rainstorm’, regn = ‘rain’, skúr = ‘shower’, veðr = ‘weather; wind’, etc. At the same
time, sverð can vary with other sword-heiti, and also within a broader equivalence class with other weapon-heiti, and still more generally with heiti for
implements of battle. This potential for substitution between individual heiti
is only within a two-element kenning, which can be called a basic kenning or
a simple kenning. To complicate matters, especially the NP of the determinant
could be realised through another kenning, and the determinant NP of that
kenning could be realised through yet another kenning, and so on, turning the
basic kenning into a complex kenning (i.e. [[NPn-GEN NP3]-GEN NP2]-GEN
NP1 = NP4, etc.). This generative system for kennings and rich vocabulary of
heiti has enormous potential for lexical variation.
There is a tremendous body of research on kennings, but almost no attention has been given to the relationship between their metrical placement and
lexical choice in realising them. The lack of attention to this topic owes in part
to the methodological problem of addressing the question (cf. Marold 1983:
43). The identification of metrically entangled kennings as established integers
of the register allows these conventional units of composition to provide a
frame for assessing lexical choices, which is the focus here. The pilot study (Frog
On the dróttkvætt metre, syntax and structuring of stanzas, see further Kuhn 1983: 33–214;
Árnason 1991: 81–148; Gade 1995: 1–72; on complexity in skaldic composition, see Wills 2009.
3
44
Frog
forthcoming) was developed on the hypothesis that kennings could become
entangled with the metre beneath a surface of variation enabled by heiti. In
other words, the pilot study set out to test whether a kenning could be “regularly used, under the same metrical conditions, to express a particular essential
idea” although, unlike Parry’s formulae, the lexical realisation of that kenning
could vary according to the phonic/lexical context of the verse in which it was
used. Lord (1960: 48–53) observed that a formula could maintain “the same
essential meaning and metrical value” although lexical choices for one element
of the formula varied according to the “acoustical context” (1960: 53; cf. Foley
1996: 19n.17). He illustrated this with a formula ‘to mount a horse’, in which a
variety of three-syllable poetic terms was used for the component ‘horse’ (Lord
1960: 48–53). Rather than one element being fixed and the other being in pragmatic variation according to the phonic/lexical environment, the pilot study
tested whether both elements of a kenning could exhibit this type of variation.
A test-corpus of battle-kennings was developed and the metrical positions
of the words constituting each kenning were mapped within a line or across
lines of a half-stanza. Mapping is done by representing each six-position line as
a numerical sequence 123456. Positions filled with a kenning’s base-word are
replaced with an ‘X’ and those of the determinant with a ‘Y’ (as well as ‘Z’ used
to represent the determinant in the determinant kenning) and ‘p’ indicating a
preposition (placed in parentheses if optional). Uppercase characters represent
the stressed onset syllable and lowercase characters represent unstressed positions. Thus, the line menn at vápna sennu = ‘men at a flyting (argument) of
weapons’ can be mapped 1-at-vápna-sennu, 1p-vápna-Xx, 1(p)YyXx, 12YyXx,
etc. Mapping the kennings in this way distinguishes them according to individual metric-structural types described by these codes. This process revealed
that more than 70% of the 340 examples in the test corpus were accounted for
by only 10 ‘basic types’.4 Rather surprisingly, the majority of battle-kennings
in the pilot study seemed potentially based on conventional models.
The first article of the present series discussed examples of the metric-structural basic type 12(p)XYy as a foundation for considering metrically entangled
semantic formulae. This metric-structural type was illustrative because the
majority of examples exhibited only two base-words which belonged to the
same semantic class of heiti (dynr = ‘din’ and gnýr = ‘roar’ as terms for ‘noise’).5
Whereas ‘complex types’ accounted for all elements in complex kennings, these were
observed to generally reflect basic types in combination that could be viewed as systematic
expansions varying a ‘basic type’.
4
E.g. Eiríkr í dyn geira = ‘...in the din of spears’ (HSt Rst 22I.6); lands folk í gný branda = ‘...
in the roar of brands (swords)’ (Ótt Hfl 9I.6).
5
Mythological Names and dróttkvætt Formulae II
45
Examples with each of these base-words were directly comparable to the formulae discussed by Lord: two elements realised a consistent unit of meaning;
one element was stable while the other element varied with a clear connection
to realising certain sounds rather than others in the line (especially for rhyme).
Unlike the formula described by Lord, the variable element was not limited
to a single semantic field. In other words, variation did not remain within a
single semantic equivalence class of heiti; the varying element only needed to
be a term which could realise the kenning as a unit meaning ‘battle’. Dominant
models for approaching kennings focus on the semantic categories of kenning
constituents and their relationship to one another (e.g. Meissner 1921; Marold
1983: 24–36). This had led to the expectation that semantic categories would
provide the basis for formulaic use of kennings in dróttkvætt (Frog, forthcoming). However, as James Fox (1977: 72) has emphasised, formal categorisation
and typology based on grouping outcomes of oral-poetic expression do not
necessarily render an accurate picture of how language functions and varies in
practice (cf. Foley 1999: 66–83). The data supports viewing metrically entangled kennings as established semantic integers of the register – as a variety
of formula. On the other hand, variation could not be reduced to synonymsubstitution within a semantic equivalence class. This yields information about
paradigmatic substitution in kenning production. When this is observed in the
semantic integers of the tradition, it has implications for how language is functioning within that integer in the production of verses. Generally speaking, the
primary determinant on lexical choice appears to be pragmatic, based on metrical viability and the phonic/lexical context of the specific verse; the semantic
category of the heiti needed only to be viable for forming the battle-kenning.
This observation in determinants was paralleled by observing corresponding variation in base-words when additional examples of the basic type were
considered.6 In other words, the kenning was not formed in a hierarchical
process of determining a semantic combination like NOISE of IMPLEMENTS
OF BATTLE → BATTLE and selecting appropriate heiti; the semantic integer
BATTLE was the basis and the specific semantic combination forming the
kenning was an outcome of pragmatic variation according to the context. In
the generative production of kennings, this suggests that lexical choice did not
necessarily require reflection on the specific semantic category of heiti. A poet
fluent in the idiom could complete this unit on the basis of phonic needs (e.g.
rhyme or alliteration) without having to reflect on whether the determinant
In basic type 12(p)XYy battle-kennings, this was particularly apparent because base-words
other than dynr and gnýr were found almost exclusively where the base-word was required to
carry alliteration.
6
46
Frog
was a valkyrie-heiti or a spear-heiti.7 In other words, this unit BATTLE could
be used in composition “without recourse to any form-meaning matching of
any sub-parts it may have” (Wray 2008: 12).
Poetic equivalence vocabulary, as in an equivalence class of heiti, may be
compared to synonymy in aesthetically unmarked discourse. A crucial difference is that this equivalence vocabulary develops to enable pragmatic variation
according to the phonetic and lexical environment of poetic expression. As a
historical process, oral-poetic registers develop a lexicon to ‘say the same thing’
within different verse contexts (e.g. meeting h-alliteration, a particular syllabic
structure or avoiding lexical repetition in parallelism). In some cases, this may
be constituted of alternative historical or dialectal forms that have produced
phonological allomorphs,8 as well as parallel inflections as allomorphs.9 In other
cases, words from other dialects or other languages are assimilated as poetic
equivalents (Foley 1996: 27–37; Fox 2014: 374–383) and words from other contexts may have their semantics ‘bent’ or ‘stretched’ to conform to the needs of
the poetic register (Roper 2012).10 Within poetic discourse, many of these exhibit
the potential to function as suppletive allomorphs (cf. English go / went; good /
better). Whereas allomorph variation in aesthetically unmarked discourse tends
to be conditioned by adjacent sounds (especially phonological allomorphs) and
grammatical contexts, allomorphy in poetic discourse is further conditioned by
broader phonological, lexical and metrical environments. Where variation can
be observed between lexical items within conventionalised poetic structures, this
offers information about categories and degrees of equivalence (cf. Jakobson 1987
[1956]: 111). From this perspective, the different three-syllable words for horse
in the formula ‘to mount a horse’ described by Lord can be viewed as allomorphs
in variation insofar as alternation between terms is dependent on contextual factors. Although it cannot be assumed that all heiti of a semantic equivalence class
functioned as allomorphs or that they did so in all contexts, metrically entangled
kennings suggest that, within these formulae, appropriate heiti were employed
Competent users of a poetic register can be expected to work within a metre to some degree
unconsciously as a function of internalising the register with its strategies and solutions for
producing metrically well-formed expressions. It is common that skilled performers cannot
articulate such processes better than, for example, simply saying “that the words draw one
another up that way” (Lönnrot 1845: 36).
7
Cf. Latin egŏ / egō = ‘I’ etc. based on historical change (Coleman 1999: 37–38); South Slavic
dēte / dijete = ‘child’ etc. as alternative dialectal forms (Foley 1996: 28).
8
9
Cf. Latin variants of the third person plural perfect -ēre, -ĕrunt and -ērunt (Coleman 1999:
44); Old Norse parallel genitive forms in Ygg-s and Ygg-jar = ‘Odin-GEN’.
Cf. Middle English tolk = lit. ‘translator’, poetic ‘man’ for t-alliteration (Roper 2012: 89).
10
Mythological Names and dróttkvætt Formulae II
47
as suppletive allomorphs that were characterised by semantic function within
the rhetorical figure rather than by their individual semantic value (i.e. NP1 =
[battle-kenning base-word]; NP2 = [battle-kenning determinant]). Such a kenning can be described as a semantic formula: a morpheme-equivalent integer of
the register in which both elements can be functional allomorphs. As formulae,
such kennings had potential to be produced and interpreted without reflection
and analysis by individuals fluent in the idiom: the use/recognition of appropriate
elements (heiti) in appropriate metrical positions could itself enable the apprehension of the referent without necessarily requiring the literal interpretation of
elements and parsing their relation.11 From this model of a semantic formula,
basic type 1(p)YyXx can be addressed to look at the entanglement of the lexicon
in realising a semantic formula of this type.
Basic Type 1(p)YyXx
Battle kennings of basic type 1(p)YyXx are presented here in 80 examples (noting that the complexity of skaldic verses inevitably leaves some examples open
to alternative interpretations). Examples listed below exclude cases of extended
battle-kennings in variations on the basic type. Examples in which an adjective complemented the kenning without otherwise impacting its semantics or
the metrical distribution of the determinant and the base-word are included
without special indication.12 In this basic type, the base-words are in most
This model is not intended to uniformly reduce such kennings to a purely mechanical
process. The intention here is to draw attention certain ways that kennings appear to have
functioned in the poetic tradition. The intuitive aspect of processing the poetry could also
have provided a resource for poets. For example, the line ósvífr Kraka drífu (Grani Har 1II.2)
‘...snowstorm of Kraki (legendary king)’ would appear initially as a battle-kenning (cf. 2a–b
below). This becomes the determinant for a valkyrie-name in the following line (Hlǫkk í harða
þjokkum), leading it to be reinterpreted as a silver/gold kenning (cf. Meissner 1921: 224, 228) to
form a woman-kenning (VALKYRIE OF GOLD). In this case, the semantic play with a battlekenning is likely strategic: the verse states that this woman will not stop weeping owing to the
aggressive deeds of the king being praised.
11
E.g. Anon Liðs 4I.7–8, Svert Lv 2IV.3–4 and Edáð Banddr 5I.5–6 – N.B. that the use of
adjectives may nevertheless be syntactically important, e.g. for forming a prepositional phrase
in which the preposition is in a different line, as in ESk Ingdr 4II.3–4 and HólmgB Lv 2V.3–4.
Citations of verses are by sigla and stanza numbering of the Skaldic Database. Citations are with
reference to the published editions of the associated Skaldic Poetry of the Scandinavian Middle
Ages edition where these were available and with reference to Finnur Jónsson’s critical edition
(1967) where they were not.
12
48
Frog
cases equivalent to the determinants in terms of syllabic quantity, yet the inversion of word order to 1(p)XxYy – the word order conventional to type 12(p)
XYy – is almost never found.13 The base-word of a 1(p)YyXx kenning is in the
position to carry rhyme in the line, and the examples foreground evidence
of conventionalised rhyme-pairs in realisations of this basic type. Evidence
of rhyme-pairs can then be looked at in relation to evidence of variation in
determinants with special attention to uses of proper names.
Of the 80 examples, 25 or 31% employ a proper name as a determinant. This
proportion is notably higher than in 12(p)XYy, which exhibited approximately
one in six or a bit more than 16%, and this relative frequency dropped to one in
eleven or about 9% if the crystallised expression gnýr Gunnar = ‘roar of Gunnr’
is counted only once. As with 12(p)XYy, metrical entanglement of the lexicon
is also evident in this case. For example, 12 of these 80 examples (15%) present
vápna = ‘of weapons’ as the determinant (9 of 12 preceded by a preposition).
This frequency is more striking in light of the pilot study, where vápn appeared
as a high-frequency determinant in the general statistics, yet its use appeared
connected to only two metric-structural types. This makes it reasonable to
consider use of vápn in 1(p)YyXx to be conventionalised. However, the present
case study focuses on the possibility that whole semantic categories of heiti
may become metrically entangled as preferred word-choices for co-occurrence
with certain base-words. It presents evidence that, when realising basic type
1(p)YyXx battle-kennings, the selection of base-word-determinant combinations does not appear free in all cases. Certain base-words predominantly
(but not necessarily exclusively) co-occur with determinants from a particular
equivalence class. This is interpreted as reflecting an indexical association
The marked exception is uses of the base-word lexical set þrima/þryma/þruma = ‘thunder,
noise’ that in the pilot study was found exclusively in 12Xx56 constructions, suggesting general
metrical entanglement of the lexical item(s). This base-word accounted for all but one example
of generative constructions in basic types 12XxYy and YyXx56 (8 examples total), with the implication that these were related to the metrical entanglement of þrima/þruma/þryma rather than
being freely generative. The only other example that I have presently identified of the inverted
word-order 1(p)XxYy is allprútt, éla Þróttar (HSt Rst 2I.7). It is also possible to approach varat
of-byrjar ǫrva (Eskál Vell 7I.1) and norðr – glym-hríðar borða (ÞjóðA Lv 3II.6) as variations on
a 12XxYy construction by regarding the element in position 2 as prefixed to the base-word for
the metrical completion of the line. This element is unnecessary for the battle-kenning (of- is
an intensifier whereas glymr = ‘crash’ can function as an independent base-word in battle-kennings – i.e. 12Xx56 → 1xXx56 and 1X2X1x56). However, this sort of compounding is particularly
difficult to analyse on the basis of two examples when it can otherwise appear possible to view
these as variations on a basic type 1XxxYy or 1X34Yy. Nevertheless, the examples appear so
unusual in the dataset of the pilot-study that it is not clear that they should be regarded as
reflecting a variation on a basic type as a semantic formula per se.
13
Mythological Names and dróttkvætt Formulae II
49
or link between these base-words and a broad semantic equivalence class of
heiti as determinants. This indexical linkage of categories of language in use
of the register would constitute a historically structured “synchronic stylistic
habit” (Foley 1993 [1990]: 192–194) that was internalised by individuals with
the poetic system.14 The indexical association of a whole class of lexical items
for completing a formula would be directly comparable to the conventional
completion of a particular South-Slavic epic formula with ‘Turkicisms’, as discussed by John Miles Foley.15
The different metrical environments of odd lines and even lines condition
variation in the realisations of type 1(p)YyXx battle-kennings. In odd lines,
the kenning determinant always carries alliteration, which can be considered
an integrated feature of the formula. There seems to be a general tendency
for alliteration to be carried by the determinant and the word at the onset
of the line. However, this remains a tendency and does not appear to be a
prescriptive feature of formula use – in contrast to the regularity of baseword-determinant alliteration in odd-line use of 12(p)XYy battle-kennings
(Frog 2014a: 126–129). Rhyme is much more regular (although not found in
In other words, the process of becoming competent in the poetic idiom did not simply
involve internalising individual words and formulae as abstract constituents of the register.
These elements were internalised in relation to situated uses, and the patterns of how and where
these were used became (unconsciously) internalised with them. The limited range of uses of a
word like vápn in battle-kennings is not related to metrical constraints: poets simply did not use
it in other contexts, whether because they simply ‘did not think of it’ when forming other basic
types of kennings or this heiti ‘just did not sound right’ in other contexts. However, when poets
formed a NP2-GEN NP1 battle-kenning of basic type 1(p)YyXx or a NP2-NP1 battle-kenning of
type YX3456, vápn ‘came to mind’. This is an indicator that the internalisation of the formulae
associated with these basic types involved the internalisation of an indexical association with
vápn as a determinant – it became a ‘natural’ word-choice. The discussion below will focus on
similar indices that develop between a base-word used in the basic type and the words with
which it co-occurs.
14
Following Foley, the verb učiniti = ‘to do, make’ is a Turkish loan used within the formula
[x] učin-ijo/-ili/-ila/-iti/etc., where the [x] is a two-syllable object of the verb and the formula
forms the second colon of an epic line. Of the 164 examples of this formula surveyed by Foley,
145 were completed by Turkish loans while only 19 were completed by a vernacular term. This
provides evidence that the formula was internalised with an indexical association with ‘Turkicisms’ as a category for completing the formulaic unit. (Foley 1993 [1990]: 192–194.) It may
be pointed out that this does not mean that singers were conscious of učiniti as a Turkish loan
(which may be contested) or even of the elements completing the line as Turkish loans (any
more than an English speaker would be conscious of a Latinate versus a Germanic etymology in
the background of preferred lexical choices in certain registers). Instead ‒ and indeed however
its background might be historically reconstructed ‒ this is an illustration of the conventionalisation of completing a formula within a broad category of lexical items.
15
50
Frog
every line):16 rhyme is almost always carried by the base-word and a syllable
in one of the first two positions in the line. Two of the exceptions may be
attributed to metrical variation in which the first and third positions of the
line carry rhyme and the fifth does not (4a.iv, 6.iii). The latter of these is from
Snorri Sturluson’s (1178–1241) illustration of what he describes as a distinct
verse form called Fleins háttr = ‘Fleinn’s verse-form’ in his ars poetica known
as Edda (Háttatal 57, Faulkes 1999: 25).17 Rhyme between the base-word and
determinant only occurs in four instances, in which both elements also carry
alliteration: Viðris veðr = ‘weather of Viðrir (Odin)’ (3b.iii), Ilmar jalmr =
‘yammer of Ilmr (valkyrie)’ (5b.ix) and two examples of Ala él = ‘squall of Áli
(hero)’ (8.ii–iii). It is striking that all four have proper names as determinants.
It is also striking that 10 (77%) of the 13 examples of this basic type in which
alliteration is carried by both parts of the kenning have a proper name as a
determinant (3b.3–4, 4b.ii–iii, 5b.vii, 5b.ix, 8.ii–v; cf. 1b.ii, 7.vi, 8.vi). However,
the number of examples remains too limited to draw any conclusions. This
observation will be therefore be left aside and returned to in the third part of
this series, where attention will turn to patterns of preferred semantic categories associated generally with a metric-structural type.
In even lines, the first (stressed) syllable in the line carries alliteration with
the preceding line while neither element in the kenning can alliterate with
this syllable or with each other. In these lines, rhyme is invariably distributed
between the base-word at the end of the line and one of the first two metrical positions (otherwise it is absent). This means that the determinant in
Rhyme may be imperfect in 3a.iii (-eð- rhymed with -eðr-). Some of the examples which
might be described as lacking a rhyme exhibit a subtle rhyme between the onset syllable and
following preposition or other light part of speech (3b.1, 3b.v, and possibly also 4b.i–ii, 8.iv).
Although equivocal, this could potentially be an alternative strategy to conventional rhyme for
integrating lines into the preferred acoustic texture of the poem (cf. Frog & Stepanova 2011:
201; Frog 2014b: 19–20).
16
Snorri formally differentiates this and certain other metrical variations as verse forms
distinct from the dróttkvætt metre proper. These metres are treated here as dróttkvætt because
they are unambiguously variations of the basic dróttkvætt metre. It should be noted that, on
the one hand, Snorri’s distinction of ‘verse forms’ was not grounded strictly in terms of metrics
in the modern sense, and on the other, he was systematising and formalising skaldic poetics.
His tour de force of illustrating and explicating more than 90 verse forms was in many cases
built on systematising a variation for composing a couplet as a basis for a complete, eight-line
stanza. In addition, Snorri’s differentiation between what he views as acceptable and what ‘old
poets’ viewed as acceptable suggests that Snorri’s differentiation of certain metrical and stylistic variations as ‘not dróttkvætt’ cannot be assumed to represent a uniformly conventional
view held by all poets in all periods, or even to represent all poets in 13th century Iceland. (See
Faulkes 1999.)
17
Mythological Names and dróttkvætt Formulae II
51
1(p)YyXx battle-kennings never carries rhyme or alliteration in even lines.
Lexical choice in this position is therefore conditioned by syllabic but not
by phonic constraints. This was also observed in even-line 12(p)XYy battlekennings. In that case, heiti filling the position in the middle of the line were
not prompted to vary in relation to the surrounding phonic environment
(except insofar as extra alliteration or rhyme in the line should be avoided).
This allowed lexical items to crystallise as preferred elements for realising
battle-kennings of this metric-structural type, and enabled the emergence
of dynr = ‘din’ and gnýr = ‘roar’ in lexically crystallised formulae within the
12(p)XYy basic type battle-kennings (Frog 2014a: esp. 122–124). The corresponding conditions in basic type 1(p)YyXx make it is less surprising that
vápn is found as the determinant in 9 of the 30 even-line examples, or in
approximately one in every three.
Inclinations to Crystallisation in 1(p)YyXx
Example set (1a–b) presents realisations of 1(p)YyXx with the base-word
senna = ‘flyting, insult exchange’. These offer a pronounced illustration of how
the formula is realised in the differing metrical environments of even and odd
lines. The even-line variants exhibit marked crystallisation, with 4 cases of
the collocation of the base-word and the determinant (vápn), and 4 overlapping cases of the collocation of the base-word and the rhyme-word (maðr =
‘man’) plus a preposition. Together, these realise the same full line in three of
the cases. In contrast, uses of the same base-word in odd lines appear to be
in free variation.
(1a) 1pYy-sennu in even lines18
1(p)
menn at
menn at
menn at
enn til
menn at18
Yy Xx
vápna sennu
vápna sennu
vápna sennu
vápna sennu
odda sennu
Translation of (p)YyXx
‘...at the flyting of weapons’
‘...at the flyting of weapons’
‘...at the flyting of weapons’
‘...to the flyting of weapons’
‘...at the flyting of points’
Source
Hfr ErfÓl 3I.2
ÞjóðA Magnfl 9II.6
Grett Lv 36V.4
Skarp Lv 9V.2
Anon Krm 17VIII.4
The manuscript variation mann í = ‘man in’ changes the plural to a singular with skothending
rhyme rather than the aðalhending rhyme.
18
52
Frog
(1b) 1pYy-sennu in odd lines
1(p)
Gunnr um
linnr kná
enn at
enn mun
mins at
Yy Xx
geira sennu
sverða sennu
eggja sennu
ǫrva sennu
malma sennu
Translation of (p)YyXx
‘...during the flyting of spears1’
‘...the flyting of swords1’
‘...at the flyting of blades’
‘...the flyting of arrows’
‘...at the flyting of metals’
Source
Skarp Lv 11V.3
SnSt Ht 6III.5
Tindr Lv 1V.5
Þtréf Lv 1IV.5
Hfr Lv 23V.3
A corresponding but more flexibly realised pattern can be seen in examples
with the base-word drífa = ‘snowstorm’ (2a–b). In these, drífa rhymes with líf
in 4 of 5 odd-line examples. A dash indicates a syntactic break – i.e. that different parts of the line belong to separate syntactic statements. This must be
stressed because a syntactic break indicates that the collocation líf–drífa is not
associated with realising a single, coherent semantic unit. It therefore does not
qualify as a formula according to the approach used here.
(2a) 12Yy-drífu in even lines19
1(p)
gjǫfrífr –
lífkǫld
mitt líf –
ǫrt líf
líf sitt –
Yy Xx
Háars drífu
Háars19 drífu
Heðins drífu
Sigars drífu
boga drífu
Translation of (p)YyXx
‘...the snowstorm of Hárr (Odin)’
‘...the snowstorm of Hárr (Odin)’
‘...the snowstorm of Heðinn (king)’
‘...the snowstorm of Sigarr (king)’
‘...the snowstorm of bows’
Source
Refr Frag 1III.2
Eskál Vell 10I.8
VGl Lv 11V.4
GunnHám Lv 6V.8
ESk Ingdr 4II.3–4
(2b) 12Yy-drífu in odd lines
1(p) Yy Xx
Translation of (p)YyXx
Source
hæfr at Hlakkar drífu ‘...the snowstorm of Hlǫkk (valkyrie)’ HólmgB Lv 6V.3
áðr í ǫrva drífu
‘...the snowstorm of arrows’
Bjbp Jóms 30I.5
The líf–drífa rhyme collocation should be seen as a compositional resource
that is complementary to the battle-kenning as a metrically entangled semantic
formula (cf. Frog 2009: 233–239). This rhyme collocation itself appears to be
metrically entangled with the 1(p)YyXx formula, so that when the formula is
realised with drífa it is already equipped with a rhyme-word for realising a
metrically well-formed line (noting also that líf is not found independent of
Manuscripts also have Haralds in this position, which is metrically acceptable although odd
as a determinant in kennings.
19
Mythological Names and dróttkvætt Formulae II
53
drífa in any identified lines with 1(p)YyXx battle-kennings).20 Uses of dynr and
gnýr in basic type 12(p)XYy similarly combine with the almr–hjalmr–malmr
conventional rhyme-set (cf. Frog 2014a: examples 1a.xi–xiv and 2a.x–xi) to
form a verbal compositional system that I have elsewhere discussed as a ‘multiform’ for realising metrically well-formed lines (see Frog 2009: 239–243).21
The pilot study on the metric-structural types of battle-kennings revealed that
the metrical entanglement of rhyme-pairs in this way was not exceptional, and
that rhyme-pairs can exhibit collocations exclusively in relation to a metricstructural type (Frog forthcoming).
In contrast to examples with senna (1a–b), 5 of the 7 examples with drífa
employ a personal name as a determinant: twice the Odin-name Hárr = ‘High
One’, once the valkyrie-name Hlǫkk and the names of two mytho-heroic kings,
Heðinn and Sigarr. In even lines, the pattern of distribution between the 4
examples of the rhyme collocation (líf–drífa) and the 4 uses of a personal
name determinant across 5 examples parallels the distribution of recurrent
lexical material in examples with senna (menn–senna and vápna–senna). The
predominance of personal names here is more striking because word-choice
of the determinant in even lines is not conditioned by meeting alliteration or
rhyme. In other words, when these lines present the same conditions that can
enable lexical stability in the role of vápn above, the predominance of 80%
of determinants as personal names is a striking anomaly. In accordance with
Within the pilot study, the líf–drífa collocation was found in two of the three examples of
drífa in this basic type and in two of the three examples of drífa’s use in compounds of basic
type YXx456. This collocation accounted for 4 of the 5 cases in which drífa carried rhyme in
these two basic types. Within the pilot study, drífa was attested in 10 examples across 5 basic
types and carried rhyme in 9 of these instances.
20
‘Multiform’ was coined as a formally defined technical term by Lauri and Anneli Honko
(1995; 1998; see also Honko 1995; 1998: 100–116; 2003: 113–122) in order to describe systems
of formulaic expressions and words that are associated in the memory of an individual singer
and that provide a flexible framework for producing expression. The multiform is considered
to function as a system at the level of verbalisation or texture rather than necessarily realising
a specific unit of content or meaning. This initial approach to multiforms was concerned with
particular questions about flexibility in epic reproduction that had not yet been sufficiently
explained through Oral-Formulaic Theory. The Honkos were not concerned with the phenomenon of multiforms per se. In several articles on different oral poetries, I have significantly
developed and refined the approach to multiforms, introducing the concept of equivalence
classes, metrical and semantic conditioning in lexical variation, syntax and also typologies, as
well as situating multiforms systematically in relation to linguistic formulae defined in terms
of morpheme-equivalent units (Frog 2009; 2010; 2011; 2012; cf. also Drout 2011). The dynamics of multiforms in skaldic dróttkvætt composition may be exceptional in their complexity in
proportion to the size of the unit they are conditioned to realise.
21
54
Frog
conventions of this basic type of battle-kenning, the personal name carries
alliteration in the odd-line example, but the frequency of personal names still
appears high.22 The lack of use of personal names as a determinant for senna
may be connected to a broader pattern of use of this term as a base-word in the
register: according to Rudolf Meissner’s (1921: 198) fairly comprehensive survey of kennings in the skaldic corpus (including all metres), personal names
are not attested as a determinant for senna-based battle-kennings. Meissner’s
(1921: 178) examples for drífa present a different picture: of 14 simple or basic
battle kennings (i.e. with not more than two elements), 6 or just under half
have a proper name as a determinant. However, half of Meissner’s examples are
12Yy-drífu battle-kennings, without which only 1 of the 7 remaining examples
have a proper name as a determinant. Against the background of Meissner’s
survey, the use of proper names in 12Yy-drífu battle-kennings appears quite
prominent, as does the proportionate number of drífa-based battle kennings
of the metric-structural type. This could be a statistical anomaly, yet a corresponding pattern emerges in uses with veðr = ‘weather’ as a base-word:
(3a) 1(p)Yy-veðri in even lines
1(p) Yy Xx
Jóans feðr Hnikars veðri
hleðr í Gunnar veðri
heðan í róstu veðri
Translation of (p)YyXx
‘...the weather of Hnikarr (Odin)’
‘...in the weather of Gunnr
(valkyrie)’
‘...in the weather of a riot’
Source
Svert Lv 2IV.3–4
Þfagr Sveinn 5II.4
SvB Lv 3V.2
(3b) 1(p)Yy-veðr-INFL in odd lines
1(p) Yy Xx
hvar er í Hildar veðri
heiðr at Hildar veðri
Translation of (p)YyXx
Source
‘...in the weather of Hildr (valkyrie)’ Grett Lv 14V.3
‘...at the weather of Hildr (valkyrie)’ GDrop Lv 4V.7
It may also be worth observing the line vífs gørninga drífu (Bbreiðv Lv 6V.8) = ‘sleet of the
wife of sorcery’, where the rhyme with drífa in the line is accomplished with the base-word of a
valkyrie-kenning (i.e. equivalent to Hlǫkk), which is itself the determinant for line-final drífa in
the formation of the battle-kenning. This is a complex kenning that is not a variation on basic
type 1(p)YyXx. Although analysable as basic type Y234Xx + XYyy56 = YZzzXx, neither basic
type would be a conventional basic type for battle-kennings, and the occurrence of the baseword in positions 5–6 appears generally exceptional for battle-kennings. Future studies of the
different basic types of a semantic formula will make it possible to assess the degree to which
the sort of indexical associations addressed here are specific to a metric-structural type or may
be shared across types in which e.g. the base-word occurs in particular metrical positions (e.g.
when Xx = positions 5–6).
22
Mythological Names and dróttkvætt Formulae II
Dreif at Viðris veðri
naðr í Virfils veðri
‘...at the weather of Viðrir (Odin)’
‘...in the weather of Virfill
(mythic king)’
Skýtr at Skǫglar veðri ‘...at the weather of Skǫgul
(valkyrie)’
‘...in the weather of Gǫndul
glaðr – í Gǫndlar
veðrum
(valkyrie)’
teðr – í tognings veðri ‘...in the weather of the sword3’
55
Tindr Hákdr 3I.1
Grett Hallfl 2V.3
SnSt Ht 54III.1
Eskál Vell 6I. 3
Balti Sigdr 3II.3
Here, 8 of 10 examples employ proper names: five with valkyrie-names (Hildr
twice, Gunnr, Gǫndul, Skǫgul), two with Odin-heiti (Hnikarr, Viðrir), and one
with the name of a mytho-heroic king (Virfill). The Odin-heiti also appears in
a variation in the couplet þryngr at Viðris veðri / vandar, góðr fyr hǫndum (Vígf
Lv 1I.3–4), where the complex kenning Viðris vandar veðr = ‘the storm of the
rod of Viðrir’ = ‘storm of the SWORD/SPEAR’ = BATTLE] can be seen as an
extension of Viðris veðr (3b.iii) without compromising the form or semantics of
the basic type. This type of variation was also observed in basic type 12(p)XYy
where the verbally crystallised formula Gunnar gnýr = ‘roar of Gunnr (valkyrie)’
was extended to Gunnar gagls gnýr = ‘roar of the bird of Gunnr’ = ‘roar of
the raven’ = BATTLE] (123-gný-Gunnar/gagls-23456) (Frog 2014a: 124–125).
According to Meissner’s survey, the use of proper names with veðr is much
more common: 19 of the 38 basic kennings, or 50%. This drops to 11 of 28
without examples of 1(p)YyXx battle-kennings, and thus the 80% of proper
name determinants in the type can be contrasted with 39% in all other types
and metres. The use of proper names with veðr can therefore be considered
exceptionally pronounced in basic type 1(p)YyXx. Nevertheless, the proper
names extend across semantic categories of heiti – they are not exclusively
valkyrie-names or Odin-heiti.
Infrequent Base-Words in 1(p)YyXx
This prominence in the use of proper names with drífa and veðr not only contrasts with their absence in use with senna (1a–b), but also with the use of
proper names accompanying low-frequency base-words in identified 1(p)YyXx
battle-kennings. For the present discussion, base-words will be considered lowfrequency in a type if they are found only in five examples or less. Although
7 examples of 1(p)YyXx are found with the base-word leikr = ‘play, sport,
game’, 4 of these are found in a single poem, which may skew the image of
56
Frog
its frequency. Examples with leikr are therefore listed here as potentially lowfrequency (4a–b):
(4a) 1pYy-leiki in even lines23
1(p)
ǫrn at
gaukr at
bleikr frá
hǫgg at
Yy Xx
sverða leiki
sverða leiki23
sverða leiki
eggja leiki
Translation of (p)YyXx
‘...at the play of swords1’
‘...at the play of swords1’
‘...from the play of swords1’
‘...at the play of blades’
Source
Anon Krm 22VIII.8
Anon Krm 16VIII.4
Anon (TGT) 30III.2
Þjóð Lv 1I.2
(4b) 1pYy-leiki in odd lines
1(p)
hversu at
hǫtt at
margr var at
Yy Xx
Translation of (p)YyXx
lǫgðis leiki ‘...at the play of the stabber (sword)’
Hildar leiki ‘...at the play of Hildr (valkyrie)’
Laufa leiki
‘...at the play of Laufi (sword)’
Source
Anon Krm 21VIII.7
Anon Krm 13VIII.3
Bjbp Jóms 18I.7
Proper names are found in 2 of the 7 examples with the base-word leikr = ‘play,
sport, game’ (4b.ii–iii) (cf. also Meissner 1921: 199). One is the valkyrie-name
Hildr (4b.ii) and the other is the name of a mytho-heroic sword Laufi (4b.iii).
L-alliteration is not frequent (cf. Hollmérus 1936). Word choice in the combination of Laufi and leiki has undoubtedly been conditioned to some degree
by phonic demands of alliteration. Although alliteration could be considered
a determinant in word-choice, Laufi seems always to alliterate when used
as a heiti in dróttkvætt poetry more generally while the sword-heiti lǫgðir =
‘stabber’ (4b.i) shows that equivalent terms capable of l-alliteration were also
available. It seems worth observing that 5 of the 7 examples of this basic type
are quite specifically sword-heiti while a sixth can be interpreted that way as
well (cf. English blade as a synonym for ‘sword’). It is at least possible that
semantics of an entangled equivalence class could have been involved here.
Additional examples of low-frequency base-words that are found three
times or less in this data present 25 examples (5a–b).24 This presents one case
of full-line correspondence between two poems (5a.iii–iv; N.B. the syntactic
break in the line of one but not the other), a second case of variation only in
the preposition (5a.i–ii), and a total of 7 examples employing the determinant
vápn (5a.iii–vii, 5b.ii, 5b.xiv), discussed above.
A variant of this line reads: “geto vid soknar lęte” (getu við sóknar læti).
23
3 x þing, snerra; 2 x galdr, glygg, gnaustan, hagl, mót, jalmr; 1 x brestr, byrr, flaumr, glamm,
kveðja, rǫdd, teiti.
24
Mythological Names and dróttkvætt Formulae II
57
(5a) 1(p)YyXx in with low-frequency base-words in even lines
1(p)
gagls fyr
gagls við
hjaldrs at
hjaldrs – at
framm í
geystir –
Hringr á
yrþjóð –
draum í
austr ór
hjalm at
veitk sǫnn –
ritr – at
Yy Xx
strengjar hagli
strengjar hagli
vápna galdri
vápna galdri
vápna glammi
vápna brestu
vápna þingi
Heðins byrjar
sverða flaumi
malma gnaustan
geira jalmi
Hugins teiti
hjalma móti
Translation of (p)YyXx
‘...before the hail of the string’
‘...at the hail of the string’
‘...at the chant of weapons’
‘...at the chant of weapons’
‘...at the clatter of weapons’
‘...the crash of weapons’
‘...at the assembly of weapons’
‘...the wind of Heðinn (king)’
‘...in the torrent of swords1’
‘...out of the gnash of metals’
‘...at the yammer of spears1’
‘...the joy of Huginn (raven)’
‘...at the meeting of helmets1’
Source
ESk Ingdr 3II.6
Hfr Hákdr 3III.2
ESk Geisl 43VII.2
Þmáhl Máv 10V.6
ÞjóðA Magn 7II.8
Bjbp Jóms 25I.8
Egill Lv 13V.8
Eskál Vell 21I.8
Bragi Rdr 3III.4
Hfr ErfÓl 22I.2
Arn Þorfdr 10II.2
ESk Geisl 41VII.6
Anon Krm 7VIII.8
(5b) 1(p)YyXx in low-frequency base-words in odd lines25 26 27
1(p)
Mér lízt
verr hafa
ǫrr í
sǫng at
geng ek
sleit at
þiggr at
viggs, í
búumk við
Hræddr fór
mest, í
mjǫð, fyr
Yy Xx
malma snerru
vápna snerru
odda snerru25
sverða þingi
geira þingi
sverða móti
Gǫndlar glyggvi
vápna glyggvi
Ilmar26 jalmi
hjǫrva27 raddar
malma gnaustan
malma kveðju
Translation of (p)YyXx
‘...the onslaught of metals’
‘...the onslaught of weapons’
‘...in the onslaught of points’
‘...at the assembly of swords1’
‘...the assembly of spears1’
‘...at the meeting of swords1’
‘...at the gale of Gǫndul (valkyrie)’
‘...at the gale of weapons’
‘...at the yammer of Ilmr (valkyrie)’
‘...the voice of swords2’
‘...in the gnash of metals’
‘...before the greeting of metals’
Source
HǫrðG Lv 10V.1
GSúrs Lv 19V.5
Þfagr Sveinn 2II.3
Tindr Hákdr 2I.7
HólmgB Lv 2V.3
Edáð Banddr 5I.5
SnSt Ht 59III.7
Arn Hardr 12II.7
Hróm Lv 2IV.3
Gsind Hákdr 8I.1
Hfr ErfÓl 4I.7
Sigv Nesv 7I.7
Manuscript variants also present ǫrt and snertu in this line, forming a different but correct
rhyme.
25
The manuscript variant almir = ‘elms’ is metrically viable but dissolves the kenning and
changes the sense.
26
Manuscript variants with hjarta = ‘of hearts’ are metrically acceptable but semantically
peculiar.
27
58
Frog
Proper names are found in only 4 of 25 examples that have base-words occurring three times or less (5a–b): the valkyrie-names Gǫndul (5b.iv) and Ilmr
(5b.ii), the name of a mytho-heroic king Heðinn (5a.vi), and the name of
Odin’s raven Huginn (5a.xi), which is otherwise exceptional for battle-kennings (cf. Meissner 1921: 201). The valkyrie-name or goddess-name (?) Ilmr
is both infrequent and obscure (Hopkins 2014). Although vocalic alliteration
is the most frequent type of alliteration in Germanic verse, mythic female
names capable of functioning as a valkyrie-heiti with this alliteration seem
to have been limited to only two (Price 2002: 338–340). The use of these particular names in battle-kennings seems to have been exceptional (cf. Egilsson
& Jónsson 1931: 104 s.v. ‘1. eir’, 319 s.v. ‘2. Ilmr’, and cf. 661 s.v. ‘Ǫlrún’). The
appearance of Ilmr in this case rather than the Odin-heiti Yggr (13.iv–v) or
the hero-heiti Áli (13.ii–iii) may be directly related to its function of carrying both alliteration and rhyme, in which case the lexical choice would
be determined by phonic requirements. Among examples with infrequent
referents, the variation Gríms í gǫndlar flaumi / Gefnar mák of hefna (VGl Lv
11V.7–8) can also be mentioned. This has been interpreted not as a variation
of Gǫndlar flaumr = ‘rush/eddy of Gǫndul’, which would be semantically sufficient: as an extended kenning, gǫndul shifts to function as a common noun
in a complex kenning with the Freyja-heiti Gefn in the kenning gǫndlar Gefnar
flaumr = ‘rush/eddy of the Gefn of (battle?)28’ = ‘rush/eddy of the valkyrie’. In
this case, complementing the kenning with Gefn simultaneously accomplished
alliteration with the preceding line and accomplished the rhyme in the line in
which it appeared. This is directly comparable to the cases of Viðris (vandar)
The valkyrie-name Gǫndul has the appearance of a feminine form of the obscure term gǫndull
(listed with two attestations in The Dictionary of Old Norse Prose). Gǫndull is thought to be a term
for a staff used in controlling gandr-spirits in magical practices. The valkyrie-name, however,
has been thought to derive independently from gandr; to stem from a poetic meaning ‘wolf ’
and thus refer to beasts of battle (Tolley 1995: 69–71; Price 2002: 341). Gǫndull = ‘staff, wand’
would be appropriate as a sword-heiti and could provide a base of interpretation here (cf. Price
2002: 338). However, according to the database of Eysteinn Björnsson (–2001), this would be
exceptional for a valkyrie-kenning because GODDESS OF IMPLEMENT(S) OF BATTLE appears
almost exclusively in the form of complex kennings. Moreover, the form of the word here is the
feminine gǫndul. One of the anonymous reviewers of this article made the insightful suggestion
that gǫndul functions here as a battle-heiti. This would conform to the most common pattern of
using a goddess-heiti as a base-word in a valkyrie-kenning (Björnsson –2001). This interpretation
is viable irrespective of the etymology of Gǫndul: almost all valkyrie-names can be interpreted
as battle-heiti or battle-kennings (cf. Price 2002: 338–340), and therefore the potentially opaque
gǫndul could be inferred by poets to function in the same way. On the other hand, a variation
of this type may have been formulaic (and equally so with Viðrir and Gunnr) without requiring
consideration of possible shifts in semantics of these names (cf. Frog 2014a: 125n.44).
28
Mythological Names and dróttkvætt Formulae II
59
veðr and Gunnar (gagls) gnýr mentioned above. Thus, the doubling of viable
terms for mythological beings appears directly associated with meeting metrical demands and potentially for aesthetic effect. Such examples are illustrative
of poets playing with conventionalised resources of tradition in composition.
The frequency of personal names with the low-frequency base-words is
slightly more than 16% – directly comparable to basic type 12(p)XYy battlekennings when the 5 examples of the conventionalised expression gnýr Gunnar
are included in the tally. This does not reveal a concentration that is strikingly
anomalous. The personal names are also not bound up with a specific category of being, which contrasts with examples of 12(p)XYy battle-kennings.
However, the diversity of proper names here might be compared with the
semantic fluidity between different heiti for weapons and armour as ‘implements of battle’ – i.e. ‘Odin’, ‘valkyrie’ and ‘mythic hero/king’ may all be seen
as mythic ‘agents of battle’. The use of proper names in the examples with drífa
(2a–b) and veðr (3a–b) remains striking. These are both weather-heiti, as are
glygg = ‘wind, gale’ (5b.viii–ix) and more loosely byrr = ‘fair wind, good wind
for sailing’ (5a.viii).29 If weather-heiti are separated from the lowest-frequency
terms above (5a–b), 2 of 4 of these are collocated with personal names while
only 2 in 20 are found with other semantic categories of base-words. Of the latter two, the kenning Hugins teiti is generally exceptional: the base-word teiti =
‘joy’ in ‘joy of the raven’ would normally refer to corpses on which they feed
(cf. Meissner 1921: 203).30 Although half of the personal-name determinants
appear in two of the five examples of weather-heiti, there are so few examples
of the individual base-words that this could simply be accidental.
Base-Words Entangled with Odd-Line 1(p)YyXx
This brings us to the final three groups of examples of metric-structural
type 1(p)YyXx: those with the base-words skúr = ‘shower’, hríð = ‘storm’ and
él = ‘squall’, all of which are weather-heiti. With a total of 21 examples, these
three groups together comprise more than one quarter of the examples of
this type. Whereas examples with drífa (2a–b) and veðr (3a–b) as well as
Meissner (1921: 181–182) made more sensitive distinctions of semantic categories of basewords than is done here and addresses byrr more specifically as a wind-heiti.
29
Use of teiti can be considered to function here as a metonym for battle (i.e. as the supplier of
the corpses which bring the raven joy). The use of the proper name of Odin’s raven does not meet
demands of either alliteration or rhyme and may therefore remain striking in this construction.
30
60
Frog
other terms exhibit use across odd and even lines, these last three examples appear particular to odd lines. This suggests that these base-words are
metrically entangled with this metric-structural type of battle-kenning in
a way that conditions their use to certain types of lines (or rhythms). They
may potentially reflect lexically conventionalised formulae. However, the
patterns of use could perhaps also reflect a type of preferred lexical choice
within a concentrated lexical equivalence set (cf. also the predominance of
veðr in odd lines). Use in odd lines presents the metrical requirement that two
stressed syllables in the line should carry alliteration. In identified examples,
skúr never participates in alliteration, which might be expected as this was
perceived in Old Norse poetry as requiring an sk-alliteration (distinguished
from s-alliteration). More surprising is that hríð only carries alliteration in 1
of 6 examples when h-alliteration (undistinguished from hr-alliteration) was
the most common alliteration pattern after vocalic alliteration in Old Norse
poetry (cf. Hollmérus 1936: 64, Table 3). Together, the 15 examples of these
two terms present only one proper name as a determinant. Use of proper
names with skúr in the survey of Meissner (1921: 181) show only 1 among
19 examples of basic kennings (not the same as the one below), and 6 of 67
for hríð (Meissner 1921: 180), so low frequency here is not surprising in that
respect. In contrast, the final term él carries alliteration in 5 of 6 examples
and also appears with a proper name as the determinant in 5 of 6 examples,
which is markedly higher than the 28 of 57 or ca. 50% of examples of basic
battle-kennings listed by Meissner (1921: 178–179).
(6) 1(p)Yy-skúr-INFL (odd lines only)
1(p)
Hárs við
hverr gerir
hilmir
Mér stóð
stendr af
ár til
Jarl lætr
vér hlutum
fár má
Yy Xx
Hǫgna skúrir
hjalma skúrar
hjalma skúrir
málma skúrar
stála skúrar
eggja skúrar
odda skúrar
vápna skúrir
fleina skúrar
Translation of (p)YyXx
‘...at the showers of Hǫgni (hero)’
‘...the shower of helmets1’
‘...the showers of helmets1’
‘...the shower of metals’
‘...from the shower of steels’
‘...to the shower of edges’
‘...the shower of points’
‘...the showers of weapons’
‘...the shower of shafts’
Source
Eskál Vell 34I.3
RvHbreiðm Hl 46II.3
SnSt Ht 57III.1
Grett Lv 22V.1
SnSt Ht 55III.5
Sjórs Lv 3II.3
ÞjóðA Frag 2II.1
Anon Liðs 4I.7
Ingj Brandfl 4IV.5
Mythological Names and dróttkvætt Formulae II
61
(7) 12Yy-hríð-INFL (odd lines only)31 32 33
1(p)
glaðr varð
grǫðr þvarr
Boðit hafa
Yy Xx
geira hríðar31
geira hríðar
brodda32
hríðar
bíðum brodda hríðar
áðr réð ek odda hríðar33
Njǫrðr nam hjálma hríðar
Translation of (p)YyXx
‘...the storm of spears1’
‘...the storm of spears1’
‘...the storm of spikes’
Source
Anon Krm 16VIII.3
Hfr Óldr 5I.7
HólmgB Lv 4V.1
‘...the storm of spikes’
‘...the storm of points’
‘...the storm of helmets1’
GSúrs Lv 32V.7
Þorm Lv 13IV.3
ÞormÓl Lv 1V.5
(8) 1(p)Yy-él-INFL (odd lines only)
1(p)
sól rauð
verum í
varat í
þar er í
bál rauðk
at hann í
Yy Xx
Svǫlnis éla
Ála éli
Ála éli
Yggjar éli
Yggjar éla
odda éli
Translation of (p)YyXx
‘...in the squall of Svǫlnir (Odin)’
‘...in the squall of Áli (hero)’
‘...in the squall of Áli (hero)’
‘...in the squall of Yggr (Odin)’
‘...the squall of Yggr (Odin)’
‘...in the squall of points’
Source
HSt Rst 16I.3
Gizsv Lv 1I.7
Eþver Lv 2V.3
Bjbp Jóms 29I.5
Hfr Lv 14V.7
Anon Krm 22VIII.3
The name of the mythic hero Hǫgni appears in one example with skúr (6.i),
which is interesting for not being used in alliteration with hríð. At the same
time, use of hríð in the place of skúr in examples (6.i–iii) or (6.viii) would produce over-alliteration in the line – a factor that might be related to the fact that
examples with skúr are more strongly represented in the data-set than examples with hríð or él. In contrast, él is found with the alliterating mytho-heroic
name Áli (with near-full-line correspondence) (8.ii–iii) and the Odin-heiti Yggr
(8.iv–v).34 It is also found with the Odin-heiti Svǫlnir (8.i) in the one example
where él does not carry alliteration. There is a pronounced difference found
across these three terms both in the participation in alliteration and in the use
of proper names as a determinant. This suggests conventional differences in
how these three base-words were deployed in this battle-kenning construction.
The manuscript variant reading “gera broder” (probably gera bróðir = ‘make the brother’)
appears to lack the kenning entirely.
31
The manuscript variant brynju = ‘of the armour’ has no metrical or semantic consequences
for the line.
32
The manuscript variant skúrar results in an absence of rhyme in the line.
33
Both of these can be considered relatively common combinations according to Meissner’s
survey (1921: 183).
34
62
Frog
The limitation of evidence to odd lines should not be over-interpreted
as meaning these base-words could never appear in even-line variations
of this metric-structural type. A variation of this formula with skúr fills an
even line with a complex battle-kenning in one case. This is in the line fúrs
í Þróttar skúrum (Eskál Vell 11I.6) = ‘the showers of the fire of Þróttr’ where
the determinant is also an Odin-heiti Þróttr.35 In this case, the additional element produces the kenning Þróttar fúrr = ‘flame of Odin’ = sword’ without
impacting the form or semantics of the 1(p)YyXx basic type. This example
should be considered in relation to those examples above in which an added
element to the kenning accomplishes metrical requirements without semantic
impact. The difference here is that the additional element occurs at the onset
of the same line rather than at the onset of the following line. Here, as in cases
already addressed, the mediating element carries both alliteration and aðalhending rhyme – a rhyme also dependent here on the choice of the base-word
skúr. Although this example remains outside of the data-set as a complex kenning, it nevertheless shows that these base-words could appear in even lines, at
least in variations on the basic type. The complete lack of even-line examples
outside of this variation could then simply be a natural statistical outcome of
their patterns of use, much as dynr is observed in only one odd-line example
of battle-kenning type 12(p)XYy (Frog 2014a: examples 1a–b). It is equally
possible that the base-word veðr also functioned similarly: the appearance of
3 even-line examples out of the 10 examples (3a–b) may be an outcome of the
larger number of examples (almost twice those of hríð or él) combined with a
few statistically infrequent cases showing up in the data.
Overview of Proper Names in 1(p)YyXx
The number of examples for each base-word of battle-kenning type 1(p)YyXx
is relatively few. If the kennings realised with a certain base-word are addressed
independently, then the concentration of personal names in a particular case
might seem a bit peculiar but could be dismissed as probably little more than
a statistical anomaly. When co-occurrence is surveyed across all base-words
and these are considered together, the use of personal names as a determinant
is unambiguously concentrated around certain base-words in battle-kennings
of this metric-structural type. Overall, 25 – more than 30% – of the identified
examples of type 1(p)YyXx battle-kennings have a proper name as a determinant.
The variation Þundr, also an Odin-heiti, is found in some manuscripts.
35
Mythological Names and dróttkvætt Formulae II
63
However, 18 of these 25 appear in conjunction with only 3 base-words: drífa
(2a–b), veðr (3a–b) and él (8). Co-occurrence with these three terms accounts
for 75% of the uses of proper names in battle-kennings of this basic type. The
remaining 7 proper name determinants are distributed across 18 other basewords, accounting for slightly more than 12% of the 57 examples using these
base-words. This is between the ca. 16% found with basic type 12(p)XYy, and
that type’s adjusted calculation of 9% when the 5 uses of the conventionalised
Gunnar gnýr kenning are counted only once. Although drífa, veðr and él all
belong to a common equivalence class of weather-heiti, corresponding use of
proper names is not associated with the base-words skúr and hríð, which belong
to the same equivalence class. The pattern of metrical entanglement associated
with base-words must therefore be considered to be at the level of lexical items on
an individual basis rather than at the abstract level of semantic equivalence class.
Personal names used as determinants in realising type 1(p)YyXx battlekennings do not appear to be bound to a particular category of heiti such as
‘Odin’ or ‘valkyrie’, as was the case in type 12(p)XYy battle-kennings. Instead,
personal names seem to function as a more fluid but distinct category which
includes both terms from the heroic sphere and terms for beings associated
with mythology – i.e. all proper names that can be used as simple determinants in battle-kennings with these base-words. Identifying that category as
‘grammatical’ (proper nouns) as opposed to ‘semantic’ (nameable identities)
may underestimate the degree to which heiti of this broad group was most
probably an intuitively internalised category within the register that might not
necessarily fully correspond to ontologies or grammatical categories current
outside of the register.36 The distribution of proper names as determinants in
examples of type 1(p)YyXx battle-kennings nevertheless appears to reflect the
metrical entanglement of a category of preferred determinants for a particular
set of lexical items as base-words.
The metrical entanglement of a particular category of determinants with
three base-words can be further contextualised among other patterns in the
data. The general prominence of the determinant vápn = ‘weapons’ suggests
that this determinant was metrically entangled with this battle-kenning basic
This observation carries the implication that additional lexical items may have been functionally identified with this group within the register but presently remain unrecognised by
scholarship owing to modern presumptions about categories to which terms in the register
should be assigned. It may be possible to confirm or refute this possibility through future studies
of patterns in language use, which will, however, most likely require the gradual development of
infrastructures in the form of research on different aspects of metrically contextualised language
use in dróttkvætt as a necessary foundation and context for discussion.
36
64
Frog
type. The use of vápn seems simultaneously to be a metrically entangled lexical preference at the general level (5b.ii, 5b.ix, 6.viii) of a potentially distinct
socially circulating formula (1a.i–iv, 5a.ii–vii) similar to dynr (Frog 2014a:
example set 1a–b) and gnýr (ibid.: example set 2a–b) – i.e. 1(p)-vápna-Xx.
The potential conventionalised formula vápna senna (1a.i–iv) may be a still
more crystallised realisation of this formula, a variation of it paralleling the
expression gnýr Gunnar, discussed in the first part of this study (Frog 2014a:
124–125). The apparently conventionalised use with senna in even lines highlights the metrical entanglement of the lexicon with the battle-kenning basic
type. The near-exclusive use of the base-words skúr (6), hríð (7) and él (8) in
odd lines (cf. also veðr) indicates that the use of these words is not freely generative in the realisation of type 1(p)YyXx battle-kennings. This implies that
these lexical items should be considered metrically entangled and potentially
formulaic in their use – odd-line formulae of the broader semantic battlekenning formula of type 1(p)YyXx. Such a possibility is highlighted by the
use of proper names as determinants used with él and also by the role of él in
alliteration as contrasted with uses of skúr and hríð in the same basic type. In
addition, conventionalised rhyme-collocations were clearly evident in evenline uses with senna (1a) and drífa (2a) as base-words. As with the indexical
association between a South-Slavic epic formula and “Turkicisms” discussed
by Foley (1993 [1990]: 192–194), these sorts of indices are indications that
metrical entanglement of the lexicon within the basic type has developed a
further degree of formulaicity. Thus even the small sets of examples for each
base-word offer indications of conventionalised patterns of language use.
This, however, returns us to Fox’s (1977: 72) concern over viewing language
through the lens of typology. On the basis of the preceding discussion, it is
possible to extrapolate each distinctive case of patterned use as a conventional
formula with variations that might be described as, for example:
(9) Examples of hypothetical abstract formulae from example sets of 1(p)YyXx above
(1a–b)
(2a–b)
(3a–b)
(6)
→
→
→
→
1(p)
1(p)
(líf-2/1-líf)
1(p)
12
(7)
(8)
→
→
12
1(p)
Yy
(‘weapons’)-GEN
(‘agent of battle’)
(‘agent of battle’)-GEN
(‘implements of battle’)
Xx
sennu
drífu
veðr-INFL
skúr-INFL
(odd lines only)
(‘implements of battle’) hríð-INFL
(‘agent of battle’)
él-INFL
(odd lines only)
(odd lines only)
Mythological Names and dróttkvætt Formulae II
65
The image that this produces may present a more or less accurate descriptive
abstraction of examples from the data. At the same time, this sort of description
separates these ‘formulae’ from one another leading to an inclination to isolate
them – i.e. as Formula 1, Formula 2, Formula 3, etc. This may be misleading with
regard to generative use of the oral-poetic register in the production of metrically well-formed verses. This is highlighted by the potential for 1(p)-vápna-Xx
to also be regarded as a distinct formula that intersects with the models of formulae described in (9) according to grouping items by base-words.
The crystallisation of verbal elements into distinct formulae in basic type
1(p)YyXx battle-kennings is quite possible and even probable, but this would
be only one quite narrow phenomenon in language practice. If the examples of
verbally identical and near-identical lines above are not considered accidental
(1a.i–5, 5a.i–ii, 5a.iii–iv, 8.ii–iii), then it appears that lines of socially circulating poetry could be adapted directly as a verbal template in composition. If
this is the case, then more of these examples may reflect this same strategy
although verbal variation in the adaptation – e.g. between equivalent heiti – is
less directly observable (cf. especially 3b.i–ii, 7.i–ii). ‘Paradigmatic substitution’ in such an adaptation would break down the formal criteria according to
which comparisons have been made above, potentially leaving the adaptation
more or less undetectable according to these methods (cf. Frog 2009: 240–243).
However, fluency in a register is characterised by an ability to move beyond
exemplar models (of whatever sort) toward internalised patterns of language
use abstracted from the patterned conventions of social practice (cf. Harvilahti
2000; Bybee 2006; Goldberg 2006). Rather than basic type 1(p)YyXx battlekennings being divided into a skúr-formula, a drífa-formula, a vápn-formula
and so forth, the relationships between these could potentially be extremely
fluid, metrically entangled verbal systems. In other words, by engaging the
metrically entangled basic type 1(p)YyXx as a semantic formula, a competent
user of the register would also be engaging a metrically entangled verbal system
conventional for appropriately realising that formula. Realising that formula
through language would involve lexical choices. At that point, lexical choices
such as whether to use drífa, senna or hríð as a base-word, or to use vápn, Hildr
or Yggr as a determinant would index co-occurring lexical material and patterns of use (e.g. alliteration) with which these are associated in realising the
particular formula. Those same patterns may equally involve closer and more
distant relations between alternative lexical choices among which variation
may be more natural (i.e. more likely to be intuitive and automatic). On the one
hand, this would explain variation within a semantic equivalence class of determinant linked to certain base-words. On the other hand, it can also be looked
at with regard to personal name determinants in terms of closeness of patterns
66
Frog
of use of skúr (6) and hríð (7) as opposed to él (8), as well as él and veðr (3) as
similar to drífa (2), and drífa and veðr in contrast to senna (1). At the same
time, senna and drífa can be looked at in relation to veðr, él, skúr and hríð with
regard to use in even and odd lines. To put it succinctly: internalised patterns
of language use lead to associated patterns in generative language production
within the register. The potential complexity of these processes should not be
underestimated. It should also not be underestimated that, in practice, multiple strategies and resources related to the single basic type could all function
in tandem. In other words, the sort of bottom-up generation of verses from
an abstract 1(p)YyXx semantic formula can be seen as one potential resource.
The top-down generation of verses from lexically crystallised formulae (e.g.
12Yy-hríð-INFL) could provide a distinct resource. Socially circulating lines of
poetry in which a kenning of this type was used could provide a third resource.
Rather than being independent of one another, exclusive and used in isolation
from one another, all of these could be complementary and in interaction.
Type 1(p)YyXx battle-kennings seem in general to exhibit patterns of the
metrical entanglement of the lexicon. The concentrated use of proper names as
determinants – even as a broad category – with certain base-words as opposed
to others appears to be only one aspect of that phenomenon in the production
of battle-kennings of this metric-structural basic type. This pattern suggests
that personal names had a functional role as integrated resources in composition rather than being primarily or exclusively referential when appearing in
conventional metric-structural kenning types. This supports the hypothesis
that proper names associated with mythology would be a fully integrated part
of the lexicon of the dróttkvætt register, although this does not exclude the possibility that these names could be used strategically for producing meanings
and associations in particular cases.37
References
Abrahams, Roger D. 1969 [1976]. The Complex Relations of Simple Forms. In: BenAmos, Dan (ed.), Folklore Genres. Austin: University of Texas Press, 193–214.
Agha, Asif 2007. Language and Social Relations. Cambridge: Cambridge University
Press.
Research presented here has been accomplished within the framework of the Academy
of Finland Project “Oral Poetry, Mythic Knowledge and Vernacular Imagination” of Folklore
Studies, University of Helsinki.
37
Mythological Names and dróttkvætt Formulae II
67
Árnason, Kristján 1991. The Rhythms of dróttkvætt and Other Old Icelandic Metres.
Reykjavík: University of Iceland, Institute of Linguistics.
Bybee, Joan 2006. From Usage to Grammar: The Mind’s Response to Repetition. In:
Language 82(4), 711–733.
Björnsson, Eysteinn –2001. Lexicon of Kennings and Similar Poetic Circumlocutions:
The Domain of Battle. URL: https://notendur.hi.is//~eybjorn/ugm/kennings/
kennings.html
Clunies Ross, Margaret (ed.) 2007. Poetry on Christian Subjects (2 vols.). Skaldic Poetry
of the Scandinavian Middle Ages 7. Turnhout: Brepols.
Clunies Ross, Margaret; Gade, Kari Ellen; Marold, Edith; Nordal, Guðrún; Whaley,
Diana; Wills, Tarrin 2012. General Introduction. In: Whaley, Diana (ed.) 2012.
Poetry from the Kings’ Sagas I: From Mythical Times to c. 1035 (2 vols.). Turnhout:
Brepols, xiii–xciii.
Coleman, R. G. G. 1999. Poetic Diction, Poetic Discourse and the Poetic Register. In:
Adams, James Noel; Mayer, Roland (eds.), Aspects of the Language of Latin Poetry.
New York: Oxford University Press, 21–93.
Drout, Michael D. 2011. Variation within Limits: An Evolutionary Approach to the
Structure and Dynamics of the Multiform. In: Oral Tradition 26(2), 447–474.
Egilsson, Sveinbjörn; Jónsson, Finnur 1931. Lexicon poeticum antiquæ linguæ
septentrionalis. Ordbog over det norsk-islandske skjaldesprog. 2nd edn. København:
S. I. Møllers.
Faulkes, Anthony (ed.) 1999. Snorri Sturluson, Edda: Háttatal. London: Viking Society
for Northern Research.
Foley, John Miles 1988. The Theory of Oral Composition: History and Methodology.
Bloomington: Indiana University Press.
Foley, John Miles 1991. Immanent Art: From Structure to Meaning in Traditional Oral
Epic. Bloomington: Indiana University Press.
Foley, John Miles 1993 [1990]. Traditional Oral Epic: The Odyssey, Beowulf, and the
Serbo-Croatian Return Song. Berkeley, Los Angeles and Oxford: University of
California Press.
Foley, John Miles 1995. The Singer of Tales in Performance. Bloomington: Indiana
University Press.
Foley, John Miles 1996. Guslar and Aoidos: Traditional Register in South Slavic and
Homeric Epic. In: Transactions of the American Philological Association 126, 11–41.
68
Frog
Foley, John Miles 1999. Homer’s Traditional Art. University Park: Pennsylvania
University Press.
Foley, John Miles; Ramey, Peter 2011. Oral Theory and Medieval Studies. In: Reichl,
Karl (ed.), Medieval Oral Literature. Berlin: de Gruyter, 71–102.
Fox, James J. 1977. Roman Jakobson and the Comparative Study of Parallelism.
In: Armstrong, Daniel; van Schooneveld, C. H., Roman Jakobson: Echoes of His
Scholarship. Lisse: Peter de Ridder Press, 59−90.
Frog 2009. Speech-Acts in Skaldic Verse: Genre, Formula and Improvisation. In:
Dewey, Tonya Kim; Frog (eds.), Versatility in Versification: Multidisciplinary
Approaches to Metrics. Berkeley Insights in Linguistics and Semiotics 74. New
York: Peter Lang, 223–246.
Frog 2010. Multiformit kalevalamittaisessa epiikassa. In: Knuuttila, Seppo; Piela,
Ulla; Tarkka, Lotte (eds.), Kalevalamittaisen runon tulkintoja. Helsinki: Suomen
Kirjallisuuden Seura, 91–113.
Frog 2011. Multiforms and Meaning: Playing with Referentiality in Kalevalaic Epic.
In: Huttu-Hiltunen, Pekka, et al. (eds.), Laulu kulttuurisena kommunikaationa.
Kuhmo: Juminkeko, 49–63.
Frog 2012. On the Case of Vambarljóð I: Comments on Formulaicity in the sagnakvæði.
In: RMN Newsletter 5, 22–38.
Frog 2014a. Mythological Names in dróttkvætt Formulae I: When is a Valkyrie Like a
Spear? In: Studia Metrica et Poetica 1(1), 100–139.
Frog 2014b. A Preface to Parallelism. In Frog (ed.), Parallelism in Verbal Art and
Performance: Pre-Print Papers of Pre-Print Papers of the Seminar-Workshop, 26th–27th
May 2014. Folkloristiikan toimite 21. Helsinki: Folklore Studies, University of
Helsinki, 7–28.
Frog forthcoming. Metrical Entanglement and dróttkvætt Composition: A Pilot Study
on Battle-Kennings. In: Árnason, Kristján; Aðalsteinsson, Ragnar Ingi; Eyþórsson,
Þórhallur; Carey, Stephen Mark (eds.), Approaches to Germanic Metre [working
title]. Reykjavík: University of Iceland Press.
Frog; Stepanova, Eila 2011. Alliteration in (Balto-)Finnic Languages. In: Roper,
Jonathan (ed.), Alliteration in Culture. Houndmills: Palgrave MacMillan, 195–218.
Gade, Kari Ellen 1995. The Structure of Old Norse dróttkvætt Poetry. Ithaca: Cornell
University Press.
Gade, Kari Ellen (ed.) 2009. Poetry from the Kings’ Sagas II: From c.1035 to c.1300
(2 vols.). Skaldic Poetry of the Scandinavian Middle Ages 2. Turnhout: Brepols.
Mythological Names and dróttkvætt Formulae II
69
Goldberg, Adele 2006. Constructions at Work: The Nature of Generalization in
Language. Oxford: Oxford University Press.
Hainsworth, John Bryan 1968. The Flexibility of the Homeric Formula. Oxford:
Clarendon Press.
Halliday, Michael A. K. 1978. Language as Social Semiotic: The Social Interpretation of
Language and Meaning. London: Edward Arnold.
Harvilahti, Lauri 2000. Variation and Memory. In: Honko, Lauri, Thick Corpus,
Organic Variation and Textuality in Oral Tradition. Studia Fennica Folkloristica 7.
Helsinki: Finnish Literature Society, 57–75.
Hasan, Ruqaiya 1989. Linguistics, Language, and Verbal Art. Oxford: Oxford University
Press.
Hollmérus, Ragnar 1992. Studier över allitterationen i Eddan. Helsinki: Svenska
litteratursällskapet i Finland.
Honko, Lauri; Honko, Anneli 1995. Multiforms in Epic Composition. In: XIth Congress
of the International Society for Folk-Narrative Research (ISFNR), January 6–12,
1995, Mysore, India. Papers. Mysore: Central Institute of Indian Languages, II,
207–240.
Honko, Lauri; Honko, Anneli 1998. Multiforms in Epic Composition. In: Honko,
Lauri; Handoo, Jawaharlal; Foley, John Miles, The Epic: Oral and Written. Mysore:
Central Institute of Indian Languages, 31–79.
Honko, Lauri 1995. Multiformit ja pitkän eepoksen arvoitus. In: Sananjalka 37, 117–145.
Honko, Lauri 1998. Textualizing the Siri Epic. FF Communications 264. Helsinki:
Academia Scientiarum Fennica.
Honko, Lauri 2003. The Maiden’s Death Song and The Great Wedding: Anne Vabarna’s
Oral Twin Epic Written down by A.O. Väisnen. Helsinki: Academia Scientiarum
Fennica.
Hopkins, Joseph 2014. Goddesses Unknown II: Ilmr. In: RMN Newsletter 8, 32-38.
Jakobson, Roman 1987 [1956]. Two Aspects of Language and Two Types of Aphasic
Disturbances. In: Jakobson, Roman, Language in Literature, Pomorska, Krystyna;
Rudy, Stephen (eds.). Cambridge: Harvard University Press, 95–114.
Jónsson, Finnur (ed.) 1967. Den Norsk-Islandske Skjaldedigtning A–B. København:
Rosenkilde & Bagger.
Kuhn, Hans 1983. Das Dróttkvætt. Heidelberg: Carl Winter.
70
Frog
Lord, Albert Bates 1960. The Singer of Tales. Harvard Studies in Comparative Literature
24. Cambridge: Harvard University Press.
Marold, Edith 1983. Kenningkunst: Ein Beitrag zu einer Poetik der Skaldendichtung.
Quellen und Forschungen zur Sprach- und Kulturgeschichte der germanischen
Völker, Neue Folge 80. Berlin: de Gruyter.
Meissner, Rudolf 1921. Die Kenningar der Skalden: Ein Beitrag zur skaldischen Poetik.
Bonn: Schroeder
Parry, Milman 1928. L’épithète traditionnelle dans Homère. Paris: Société d’Éditions
Les Belles Lettres.
Price, Neil S. 2002. The Viking Way: Religion and War in Late Iron Age Scandinavia.
Uppsala: Department of Archaeology and Ancient History.
Roper, Jonathan 2012. Synonymy and Rank in Alliterative Poetry. In: Sign Systems
Studies 40(1/2), 82–93.
Saussure, Ferdinand de 1967 [1916]. Cours de linguistique générale. Paris: Éditions
Payot & Rivages.
Siikala, Anna-Leena 1990 [1984]. Interpreting Oral Narrative. FF Communications
245. Helsinki: Academia Scientiarum Fennica.
Skaldic Database. URL: http://abdn.ac.uk/skaldic/db.php
Tolley, Clive 1996. Vǫrðr and Gandr: Helping Spirits in Norse Magic. In: Arkiv för
Nordisk Filologi 110, 57–75.
Wills, Tarrin 2009. The Development of Skaldic Language. In: Ney, Agneta; Williams,
Henrik; Ljungqvist, Fredrik Charpentier (eds.), Á austrvega: Saga and East
Scandinavia. Preprint Papers of the 14th International Saga Conference, Uppsala,
9th–15th August 2009. Gävle: University of Gävle, 1032–1038.
Wray, Alison 2008. Formulaic Language: Pushing the Boundaries. Oxford: Oxford
University Press.
Wray, Alison 2009. Identifying Formulaic Language: Persistent Challenges and New
Opportunities. In: Corrigan, Roberta, et al. (eds.), Formulaic Language I–II.
Typological Studies in Language 82–83. Amsterdam: John Benjamins Publishing,
I, 27–51.